Shiyang Li

Shiyang Li

PhD student at University of Minnesota-Twin Cities

Liberator: A Data Reuse Framework for Out-of-Memory Graph Computing on GPUs

Published in IEEE Transactions on Parallel and Distributed Systems (TPDS), 2023

Abstract

We present Liberator, a data reuse framework that enables efficient out-of-memory graph computing on GPUs. When graph data exceeds GPU memory capacity, Liberator intelligently partitions graph data and schedules computation to maximize data reuse across partitions, significantly reducing the overhead of CPU-GPU data transfers.

Key Contributions

  • A novel data reuse framework for out-of-memory graph processing on GPUs
  • Intelligent graph partitioning and scheduling to minimize data transfer overhead
  • Demonstrated significant speedups over existing out-of-memory GPU graph frameworks

Authors

Shiyang Li, Ruiqi Tang, Jingyu Zhu, Ziyi Zhao, Xiaoli Gong, Wenwen Wang, Jin Zhang, Pen-Chung Yew

IEEE Transactions on Parallel and Distributed Systems (TPDS), 34(6): 1954-1967, 2023.

Liberator Framework

Recommended citation: S. Li, R. Tang, J. Zhu, Z. Zhao, X. Gong, W. Wang, J. Zhang, P.-C. Yew. "Liberator: A Data Reuse Framework for Out-of-Memory Graph Computing on GPUs." IEEE Transactions on Parallel and Distributed Systems (TPDS), 34(6): 1954-1967, 2023.
Download Paper